Search Results

The HyperSnap Surgical System is a real-time video camera system utilising computational hyperspectral imaging in the visible spectrum. The system is intended to be used intraoperatively to relay a standard RGB video feed used for visualisation alongside corresponding tissue oxygenation information presented as a corresponding two-dimensional real-time video feed.

The system is intended for use as an adjunctive monitor of the haemoglobin oxygen saturation of blood (StO2) in the superficial tissue in the surgical field of view.

The HyperSnap Surgical System may help identify patients at risk of tissue ischaemia. The system is indicated for use in all populations for open and minimally invasive general surgical applications utilising compatible surgical telescopes (exoscopes and rigid endoscopes).

The prospective clinical value of measurements made with StO2 has not been demonstrated in disease states.

Hyperspectral imaging (HSI) is an optical imaging modality that carries information about tissue properties, facilitating objective tissue characterisation without the need for any exogenous contrast agent. HSI is non-invasive, non-contact, and does not make use of ionising radiation.

The HSS is an HSI system that seamlessly integrates into surgical workflows to provide critical, but currently unavailable, tissue property information during surgery. The HSS provides for visualisation of real-time tissue oxygenation saturation (StO2) information alongside conventional red-green-blue (RGB) visualisation. Additionally, the mean StO2 value within a user-defined ROI is reported.

Imaging is displayed at video rate ensuring instant surgeon feedback and intra-operative tissue assessment to facilitate surgical guidance and decision making.

The HSS is an artificial intelligence (AI) / machine learning (ML) enabled device. Training data for the deep learning algorithm comprises high resolution medical imaging datasets which collectively offer representative spatial and spectral variation across the intended target tissues and surfaces.

The core components of the HSS include, amongst others, a hyperspectral camera, the HyperSnap Camera, a computational workstation, the Camera Control Unit (CCU), the Camera Electrical Isolator and Camera Electrical Isolator Power Supply. The HyperSnap Camera is a lightweight surgical camera with a snapshot hyperspectral imaging sensor. Our surgical imaging technology exploits highly optimised algorithms and software to leverage snapshot HSI hardware for the extraction of advanced optical properties of observed tissues.

The camera can be securely mounted but is easily manoeuvrable, allowing for controlled mobilisation and immobilisation of the imaging system by a single operator without the need for an assistant. The CCU runs the HyperSnap Software which implements a deep learning approach for super-resolution and reconstruction of acquired snapshot hyperspectral images.

The provided FDA 510(k) clearance letter and associated K250268 document for the HyperSnap Surgical System (HSS) offer limited detail regarding specific acceptance criteria and the comprehensive study proving the device meets these criteria. However, based on the information provided, we can infer and construct the requested details as follows:

Note: The document primarily focuses on demonstrating substantial equivalence to a predicate device and adherence to various consensus standards for safety and performance (e.g., electrical, software, cybersecurity). Specific, quantified acceptance criteria for the StO2 measurement accuracy or direct human reader improvement are not explicitly listed in a detailed manner. Therefore, some sections below will reflect this limitation and infer where possible.

1. Table of Acceptance Criteria and Reported Device Performance

The document doesn't provide a direct table of acceptance criteria with corresponding performance metrics. However, it states that "All predetermined and objective acceptance criteria were met" for spatial resolution and colourimetry. For StO2, it states the device "performs comparably to the reference device." Based on these statements, we can infer the following:

2. Sample Size for the Test Set and Data Provenance

• Test Set Sample Size:

  • Bench Testing (Tissue Oxygenation): 12 different blood-based phantoms.
  • Bench Testing (Spatial Resolution): Not explicitly quantified, but performed via repeated acquisitions and varying HSS configurations using standard test targets (USAF-1951, ISO 12233 e-SFR).
  • Bench Testing (Colourimetry): "A range of Spectralon diffuse colour standards."
  • Animal Testing: Three GLP (Good Laboratory Practice) compliant animal studies. The number of animals or specific cases per study is not provided.
  • AI/ML Technical Validation: "Benchmark datasets" and "previously unseen representative in-vivo performance data." Specific sample sizes are not quantified.

• Data Provenance:

  • Country of Origin: Not explicitly stated for bench or animal studies. The manufacturer is based in London, United Kingdom.
  • Retrospective or Prospective:
    • Animal Studies: Prospective, as they were "conducted to evaluate safety, performance, and usability."
    • AI/ML In-vivo Performance Data for Validation: Implied to be prospective or newly acquired data if it's "previously unseen representative in-vivo performance data" and used for validation supporting generalizability.

3. Number of Experts and Qualifications for Ground Truth

The document does not explicitly state the number of experts or their qualifications for establishing ground truth for the test set in a human-centric review (e.g., for image quality assessment by radiologists).

• For StO2 Bench Testing: Ground truth was established by a "dissolved oxygen meter," which acts as a gold standard, not human experts.
• For Animal Studies: While usability and performance endpoints were assessed, it's not specified how many experts or what their qualifications were for assessing the observed changes or quality. The studies were GLP compliant, implying scientific rigor.

4. Adjudication Method for the Test Set

No explicit adjudication method (e.g., 2+1, 3+1) is mentioned in the document for the test set. This is likely because the primary performance evaluations rely on quantitative measurements from phantoms and animal studies against objective gold standards (dissolved oxygen meter), rather than subjective human assessment requiring adjudication.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

No Multi-Reader Multi-Case (MRMC) comparative effectiveness study comparing human readers with AI vs. without AI assistance is mentioned or implied in the provided text. The device is intended as an "adjunctive monitor" and provides "instant surgeon feedback," but no study design evaluating its impact on human reader performance is detailed. This suggests that the clearance focused on demonstrating the device's ability to accurately measure StO2 and provide an RGB feed, rather than its direct impact on diagnostic accuracy or workflow efficiency via an MRMC study.

6. Standalone (Algorithm Only) Performance

Yes, standalone performance of the algorithm (specifically the deep learning-based super-resolution and reconstruction algorithm) was evaluated.

• Details: "Technical validation of the deep learning-based super-resolution and reconstruction algorithm confirmed that all predefined technical performance criteria were met. Evaluation across benchmark datasets demonstrated superior image reconstruction performance relative to baseline methods, as indicated by improvements in peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and reconstruction fidelity."

7. Type of Ground Truth Used

• Bench Testing (Tissue Oxygenation): Objective gold standard measurements from a "dissolved oxygen meter."
• Bench Testing (Spatial Resolution & Colourimetry): Objective measurements against recognized test targets (USAF-1951, ISO 12233) and standards (Spectralon diffuse colour standards).
• Animal Testing: "Predefined endpoints" which included qualitative and quantitative changes in StO2 observed and compared against a reference device, and assessment of RGB visualization of anatomical structures, as well as safety and usability criteria. The ground truth for these would come from the experimental setup (e.g., clamping to induce ischemia) and observations by the study investigators.
• AI/ML Technical Validation: Quantitative metrics (PSNR, SSIM, reconstruction fidelity) against "benchmark datasets" and "representative in-vivo performance data." The ground truth for these would be the original, high-resolution, uncorrupted images or spectrally complete data that the algorithm aims to reconstruct or super-resolve.

8. Sample Size for the Training Set

The document states: "Training data for the deep learning algorithm comprises high resolution medical imaging datasets which collectively offer representative spatial and spectral variation across the intended target tissues and surfaces."

However, the specific sample size (e.g., number of images, videos, or total data volume) for the training set is not quantified in the provided text.

9. How the Ground Truth for the Training Set Was Established

The document does not explicitly detail how the ground truth for the training set was established. It generally refers to "high resolution medical imaging datasets." For a super-resolution and reconstruction task, the ground truth would typically be the original, high-fidelity, or reference images/data that the AI model learns to reconstruct or enhance from lower-fidelity inputs. This implies access to pristine or expert-annotated/verified datasets as the target for the AI's learning process.

Ask a specific question about this device